Arthur Mensch

Arthur Mensch (born 1992) is a French AI researcher and entrepreneur. After a PhD in machine learning and a stint at Google DeepMind — where he co-authored the influential “Chinchilla” paper on how to scale large language models efficiently — he co-founded Mistral AI in Paris in 2023 with Guillaume Lample and Timothée Lacroix. The startup raised one of the largest seed rounds in European history weeks after founding, and its open-weight models such as Mistral 7B and Mixtral quickly became the standard-bearers for a European alternative to the American labs.

Openness as strategy and creed

Mensch is one of the industry’s most consistent advocates for open models. He frames openness not as charity but as sovereignty: states and organizations, he argues, should own and control the AI systems that embed their knowledge and run their critical processes, rather than rent them from a foreign provider. In 2026, after Anthropic restricted model access for some foreign users, he wrote that Mistral exists so that the best AI systems stay available “outside of centralised control” — a pointed pitch for open source as the safer long-term bet.

He is equally blunt about AI doom. Mensch has repeatedly dismissed existential-risk framing as overblown and self-serving, arguing that regulation should target applications rather than the underlying models — a position that put him at odds with parts of the EU AI Act during its drafting, and with safety-focused labs whose warnings he sees as convenient moats. Critics counter that Mistral’s lobbying watered down European rules; Mensch’s reply has been that Europe cannot regulate its way to relevance without building champions of its own.

Under his leadership Mistral has partnered with Microsoft, shipped the consumer assistant Le Chat, and become the technology backbone of choice for European governments and enterprises seeking AI independence.